Morphology of an arid landscape utilising synthetic-aperture radar (SAR) and differential interferometric SAR (DInSAR), southern Riyadh, Central Arabia

This study on the southern Riyadh area examines the ERS-1/2 and ENVISAT satellites data' ability of detecting Early 10 Quaternary-Holocene trans-tensional Central Arabian graben system morphology. It also, test the hypothesis of potentially geohazardous arid region for the consequent dissolution-induced collapses and karstifications and possible recent faults reactivation. Eight Single Look Complex (SLC) amplitude images are calibrated, filtered, georeferenced, orthorectified, and filtered at a resolution of 20 metres, and compared with one another by producing 17 diachronic images of the pairs at different intervals (1996, 2003-2005, 2008). The diachronic SAR intensity imageries suggest a downthrown displacement reaching 600 15 m and eastward tilting at the bottoms of the grabens. Also, the structurally-controlled valleys are developing an eastward-running drainage system towards the oasis of Al-Kharj and capturing an older hydrologic system. Moreover, a 12-year period (19962008) of the SAR data was obtained to examine the average annual rate of southern Riyadh’s anthropological sprawl, which is estimated at approximately 390 metres/year over the 12 years and constrained by geomorphological features towards the deformed area. DInSAR imageries show the primary results obtained from the 26 May 2004 and 31 Jan. 2005 pair of images, 20 merged with 30 m resolution DEM-SRTM data for the arid region south of Riyadh to eliminate the influence of topography. DInSAR is applied in this study for its ability of detecting small displacements at the centimetre scale (1/2 wavelength). Although the DInSAR’s coherence and phase imageries suggest a fairly stable region since the last tectonic and subsequent geomorphic events, erosional and artificial changes are observed, bounded within the valleys and depressions, primarily due to aeolian and fluvial processes and agriculture. It is highly recommended to preserve the area for sustainability and economy. 25 _________

changes that had occurred in different areas such as those during a period of almost 12 years. The radar data are provided with 95 colour composition (diachronic) images that represent a 12-year period for monitoring potential hazardous features witnessing several constructions that have multiplied on the periphery of and outside the city of Riyadh. The ultimate goal of this work is to alert the community and preserve the area over the south of Riyadh and east of Al-Kharj, in Central Saudi Arabia, for future water and food security, for economic aspects, and as a national geopark for scientific research and tourism.

2 Data and methods
To fulfil the abovementioned objectives, the available radar data are acquired; two radar wave components of ERS-1/2 and ENVISAT images are found between 1996 and 2008 (Table 1). Eight amplitude images are produced, multi-dated, calibrated, filtered, georeferenced, orthorectified, and filtered at a resolution of 20 metres. These images are compared with one another by 105 producing 17 diachronic images of the pairs at different intervals (1996,2003,2004,2005,2008). The eight Single Look Complex (SLC) images of SAR satellites ERS-1/2 and ENVISAT were processed by pulse compression in the radial direction in distance and by SAR aperture synthesis in the azimuthal direction. In this type of product, each pixel is represented by a complex value (real and imaginary part) from which the amplitude and phase of the signal are extracted. For this study, eight SLC images from the ERS-1/2 and ENVISAT satellites' images have 4,904 columns and 29,715/29,708 rows for ERS-1/2 and 5,164 columns and 110 26,172, 27,314, 27,325, 27,315, 27,325, or 27,326 rows for ENVISAT. Radiometric calibration involves bringing the produced image (amplitude, intensity) to a real ground reflectance or radiance. Geo-referencing makes it possible to have the images produced in the same cartographic projection system (UTM zone 38-N, WGS84). This step was carried out automatically by assigning the geographical characteristics of each image. Orthorectification involves correcting the amplitude or intensity image to the shape of the Digital Elevation Models (DEM) from the Shuttle Radar Topography Mission (SRTM) at 30 m resolution. 115 Filtering is the removal of the speckle effect to increase the readability of an image. The data experienced radiometric calibration, processed by pulse compression in the radial, range, and azimuthal directions. These images used in this study cover an area of 100 km × 100 km with a resolution of 4 m in azimuth and 20 m in distance. In this type of product, each pixel is represented by a complex value (real and imaginary parts) from which the signal amplitude and phase are extracted. The acquisition is descending (day). The scene is illuminated to the right in the lateral view with a 23° angle of incidence in the C-band at a 120 wavelength of 5.65 cm and with vertical polarisation (V/V).
The region south of Riyadh using SAR images underwent the application of the derivative of the differential interferometry (DInSAR) technique, the coherence image, which distinguishes stable areas that have retained the phase from non-stable areas that disturb the phase. Figure 3 shows the primary results obtained from the pair of images obtained on 26 May 2004 and 31 Jan. 2005. DEM-SRTM images are used for the arid region south of Riyadh with a resolution of 30 m to eliminate 125 the influence of topography. The choice of the two dates used in the coherence image is based on feasibility conditions; data are considered by temporal baseline interval (Btemp) from one day to 12 years, adopting a spatial baseline of < 1,000 m. This study used the GAMA software, under the Linux and Windows (Cygwin) operating systems, for the realisation of the differential interferograms. All the pairs are formed from these data, and the calculated geometric characteristics of the images for each pair https://doi.org/10.5194/gi-2020-32 Preprint. Discussion started: 8 October 2020 c Author(s) 2020. CC BY 4.0 License. are processed ( Table 2). The DInSAR was performed using the Centre national d'études spatiales (CNES) method, using two 130 passes and an external DEM. The carried out treatments of differential interferograms show fringes corresponding to the phase difference. Coherence images demonstrate the reliability of the interferograms producing: (0 = minimum coherence; 1 = maximum coherence) (1) Coregistered amplitude images correspond to the realised differential interferograms and unwinding of the realised phase difference. These images are compared with one another by making colour composites at different intervals ranging from one 135 day (24 hours) to 11 years and seven months (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008); the processing steps perform and extract eight calibrated, georeferenced, orthorectified, and filtered amplitude images of the study area in the southern region of Riyadh (Fig. 4). 17 colour composition pairs are produced for different satellite intervals (ERS and ENVISAT). These intensity (colour composition) image pairs of the study area are shown in Fig. 5. The colour of each pixel of the resulting image was processed by the equation: (2) 140 in which a(p) and b(p) represent values depending on the signal strength of the pixel for the oldest and most recent component respectively; thus, pixel appears in Red [R]. Table 3 summarises the results of colour composite enhancements, while Table 4 shows the legend of the low and high signal and the reflected colour. For a pixel appearing in black and in white, signal strength is equal to: a(p) = 0, and b(p) = 1; hence C(p) = [R] (3) 145

Results
As a result of using amplitude, intensity, coherence, and interferometric phase imageries, the following landscapes were distinguished: mountainous areas, depressions and grabens, sand dune areas confirmed by the Landsat imagery (Fig. 6), and areas 150 of human activity (urbanisation, roads, and agricultural fields). The radar data have given rise to several different geomorphological phenomena, such as river captivity and the fragmentation of the surface topography of the plateaus by the grabens and depressions.

Morphology of the landscape 155
The Tuwaiq Mountain series (TM) form an overall plateau, cut through by different wadis (Fig. 1), such as Wadi Hanifah (WH), which extends from northwest to southeast, and Wadi Al Awsat (WA) and Wadi Nisah (WN), which extend from west to east and empty into a regional depression known as the Al-Kharj Depression (AD). The Wadi Hanifah and Wadi as-Silay (WS) are linear valleys running north to south, whereas Wadi Nisah and Wadi Hanifah run west to east and are originally grabens formed 160 by tectonic events (Bamousa et al., 2017;Bamousa, 2018). These wadis cross-cut steep beds between hard and brittle Mesozoic and Cenozoic clastic, carbonate, and evaporitic rocks (Fig. 2).

Amplitude and intensity images
The amplitude illustrates the specificities of the different landscapes of this region, such as the grabens and the Tuwaiq mountain range. It also shows the overlying urban extension of the city of Riyadh towards the deformed areas (Fig. 7). This extension is characterised by four directions from the southeast to the southwest, estimated at different rates, related to the 170 encountered complexity of the terrain (Table 5). In the southeast direction, the rate of extension was estimated at approximately 240 m m/year; towards the south, this was derived as approximately 460 m/year; towards the southwest, this rate was approximately 770 m/year, and towards the west-southwest, it was approximately 90 m/year. Thus, the total rate of extension is 1560 m/year, and the average of extension towards the deformed area is 390 m/year during the 12-year period. N-S profiles 1, 2, and 3, subtracted from the intensity image, show the variation of the topography from north to south (Fig. 8). It shows the effect 175 of the early Quaternary graben system and the consequent late Pleistocene karstifications that deformed the Tuwaiq Mountain series in the study area. The maximum height of elevation is approximately 1,000 m in the west, and the lowest altitude is approximately 400 m in the east, indicating a maximum 600 m of downthrown displacement and depression. Profiles 4 and 5, plotted E-W along the Awsat and Nisah valleys reveal eastward tilting and consequent capturing of the hydrologic system by the last tectonic event in the Central Arabian graben system (Fig. 8). Therefore, a new hydrologic system has developed, in which 180 Awsat is pouring into the Nisah that in turn pours into the Sahba valleys west of Al-Kharj City (Fig. 9).

Differential interferometry
In the 280-day period noted above, the southern region of Riyadh showed good coherence (yellow colour) in the high 185 relief area and poor coherence (blue-violet colour) in the low relief and agricultural areas (Fig. 9). Coherence represents the pixel stability, which distinguishes high phase values (high coherence) from those that disturb the phase (low coherence), such as vegetation and water, among others ( Fig. 9). The analysis of the landscape south of Riyadh indicates stable terrain with no landslides, subsidence, or flash floods suggested during the period 1996-2008. The consistency image made it possible to distinguish between two parts of the study area. One part retained the signal phase, showing high coherence translated by the 190 Tuwaiq mountain range; in another part, the signal phase was disturbed in e.g. the graben and valleys, indicating low coherence due to unconsolidated sediments, water, and agriculture. The produced coherence image shows the reliability of the differential interferogram performed (Fig. 9). The realised differential interferogram phase imagery does not indicate displacement fringes (Fig. 10). The course of this phase difference indicates the atmospheric phase, which correlates with the topography of the region southwest of Riyadh (Fig. 10). Altitude values (x) are not provided with a numbered scale for unknown coefficient (k) value in 195 the following equation: If (k) is equal to (1), the lowest altitude value is equal to (2π). Therefore, the interval of the provided range is equal to 50 m/colour, ranging from 400 m to 1,000 m according to the topographic map (Fig. 10).

Diachronic image interpretation
The use of the images dated 29 Sept. 2003, 12 Jan. 2004, and 26 Apr. 2004, and their colour compositions, aimed to enable observation of changes due to the autumn, winter, and spring seasons, which corresponded to the months of acquisition 205 of these three radar images. These images indicate that the observed changes were mainly due to weather conditions. The Jan data indicate a high signal strength, which can be translated into high humidity due to precipitation in the winter of 2004. In Accordingly, it can be stated that this area was morphologically stable during the 12-year period. The irregular slope factor contributed to the transport of carved materials from the highest to the lowest elevation. On the slope surface, several valleys and tributaries (Wadis: Hanifa, Nisah, Al Awsat, Laha) are present, with beds carved out of solid and fragile formations, running eastwards, forming a dense hydrographic network. Various erosion factors, such as wind and water, have contributed to the morphological evolution of the region and the recharging of near-surface aquifers in highly permeable geological formations. 220

Discussion
Active remote sensing data are among the important sources for the study and analysis of natural hazards' occurrence.
SAR is the effective, fast, and inexpensive all-weather operation data of active sensors on board orbital satellites and has the 225 ability to be enhanced by several methods. It provides two components of the backscattered signal: energy (radiometry), which is transmitted and receives good signal amplitude, and time (phase), which relies on the recorded wave distance sent by the radar sensor and the one coming from the target. Amplitude characterises the high-frequency wave images which were used in several landforms and different applications. It can be used directly as a substitute for the optical image, reflecting the surface conditions. Stable wet surfaces are represented by high radiometry values; soft arid surfaces appear with low radiometry values. Therefore, 230 radiometric corrections are very important before image visualisation and interpretation. ERS and ENVISAT are two important types of SER data, which provide valuable information with geological maps (including major geological structures such as faults, collapses, and valleys) and detect surface topography. The availability of ERS-1/2 and ENVISAT data at different dates contributes to risk management, both during and after disasters. Moreover, the diachronic analysis of these data highlights surface variations, particularly in river systems, depressions, fields, and urban areas.
Radar images can detect mass movement, such as landslides and mudflows, by combining them with DEMs, and can suggest the direction of mass movement in the region accurately. The association of the amplitude image with a DEM has shown effectiveness in determining soil surface moisture in arid and semi-arid areas (Troufleau et al., 1994). Radar images have shown their usefulness in monitoring deforestation, ice melt, and the polar environment (Budkewitsch et al., 1996a(Budkewitsch et al., , 1996b. They can be used in studying and detecting mesoscale phenomena in oceans (Laborde and Deveaux, 1996), shape recognition and land-240 use planning (Rudant et al., 1996), and crop classification (Bruniquel and Lopes, 1994;Lopes and Sery, 1997). They have their own applications in geological and structural mapping (Singhroy and Saint-Jean, 1999;Wade, et al., 2001) and detection and characterisation of hydrocarbon slicks in the seas and oceans (Mercier et al., 2004). They are very strong tools in monitoring flood extent at the time of acquisition (Brivio et al., 2002;Mcmillan et al., 2006). Moreover, they are helpful in mapping urban damage due to natural or industrial disasters and changes in coastal features (Ba et al., 2007). Synthetic-aperture radar (SAR) 245 wavelengths and polarisations provide the best results for detecting, identifying, and mapping exposed and buried channels under the sand by detecting reflectance contrast differences between the bed of buried valleys or rivers and the surrounding environment. Utility, digital processing, visualisation, and interpretation of radar data (ENVISAT, ERS 1-2) at different dates and the application of differential interference methods, such as interferometric SAR (InSAR), DInSAR, small baseline subset synthetic-aperture radar interferometry (SBAS-InSAR), and pulse-doppler radar (PS), were applied in several countries. Hachemi 250 et al. (2009;, Hachemi and Thomas, (2013), and Hachemi et al. (2014a, b;2015; applied radar data in several countries such as France, Belgium, Romania, Canada, Romania, Vietnam, Mauritania, Morocco and Algeria. Diachronic DInSAR images draped over SRTM images are the most suitable data to detect and monitor risk areas, to map them as accurately as possible, and to estimate, if necessary, the resulting damage and landscape changes. Consequently, the SAR interferometry technique has many applications in the area of deformation and geohazard monitoring, such as earthquakes, volacanoes, 255 subsidence, and landslides (Graham, 1974;Zebker and Goldstein, 1986;Goldstein et al., 1988;Massonnet et al., 1993;Fruneau and Achache, 1995;Avallone et al., 1999;Briole et al., 1999;Fruneau and Sarti, 2000;Cakir et al., 2003;Closson et al., 2003;Stramondo et al., 2005;Delacourt et al., 2007;Beauducel et al., 2000;Hachemi, 2009;Hachemi et al., 2012). DInSAR is a means of detecting small displacements at the centimetre scale (1/2 wavelength applied in this study).
This study tests the ability of the SAR and DInSAR imageries and applies them on this arid region for the reported 260 karstification (e.g. Bamousa et al., 2014) and recent activities of the Sahba fault and valley, to the east of Al-Kharj (Weigermars, 1998). The Sahba fault cuts through the biggest inland oil field (known as the Gawar oil field) and the Quaternary sand dunes, reaching the Arabian Gulf and connecting with the Zagros Mountain thrust belt. Also, with other regional Wadi Al Batin and Az-Zulfi faults forming a regional feature within Arabia, the East Arabian Block holds most of the oil and gas fields in the world (e.g. Weigermars, 1998;Bamousa et al., 2017). Therefore, this study is conducted based on three factors. Firstly, there is the 265 importance for the people of the oasis, built thousands of years ago for its mild weather, good resources of water, and agriculture, which are essential elements for living humans and animals. Secondly, the reported karstification, fault activities and the extensional geomorphology might have the potential for hazards, especially if triggered by artificial activities and expansion of the nearby cities. Finally, the study area for the surrounding oil and gas fields has its place in the economy. Moreover, this study also tests possible earthquake activities that might trigger reactivation of the faults and grabens by oil and groundwater extraction. 270

Conclusion
This study is a regional and remote sensing investigation of the morphology of an arid region, and it was possible to clearly distinguish the different aspects that characterise the area. Multi-temporal analysis, conducted using amplitude images, 275 has shown the possibility of mapping surface changes at different dates. It also shows the ability to determine and locate faults, fractures, and other geological features. It makes it possible to understand the evolution of drainage networks over different periods. These radar images address the major issue of urban sprawl, which must be reviewed and planned in a preventive manner. This research can serve as a tool for city planners and decision-makers to prepare future projects for the benefit of the environment and society. Finally, this study contributes to the development of methodologies in similar conditions, including 280 spatial and temporal identification of areas potentially exposed to natural hazards. This study recommends preserving the region of Wadi Nisah as a national geopark by the Ministry of Environment, Water and Agriculture for several reasons. In this way, all municipalities would take this area out of plans for making new residential areas to expand the two cities of Riyadh and Al-Kharj. The study area has the potential of groundwater, oil, and gas resources. The second potential is the faults' and grabens' reactivations. The third potential is that, as an oasis that has been built thousands of years ago, it can be preserved for long-285 standing food supply and therefore food safety. This study also recommends sustaining the groundwater for water safety by changing the type and style of irrigated farms to another style that consumes less groundwater.

Author contribution
The authors conceived the idea presented and contributed to the formulation of the objectives and methodology of the research, 290 in particular the writing of the initial project. The first author was involved in the administration and supervision of the project as well as in the funding acquisition and the implementation of the research schedule, leading to this publication. The second author was responsible for the acquisition of the radar images and the numerical processing of the data and statistical calculations (data curation). The third author suggested the study area for this project, and was involved in documenting the introductory parts of the manuscript for his local background. All the authors contributed to the designing and describing of the carried-out 295 methodology, the investigation processes, formal analysis; they were involved in the verification, validation and interpretation of the results and contributed to the writing of the final manuscript.

Acknowledgments
The research team would like to thank the Deanship of Scientific Research at King Abdulaziz University for its financial support 300 for this investigation . We also thank the European Space Agency (ESA) for providing us with SAR images of the ERS and ENVISAT satellites as part of a research project (No. 28248). The work is based on the exchange of experience between three universities (the Department of Geography and Geographic Information Systems, King Abdulaziz https://doi.org/10.5194/gi-2020-32 Preprint. Discussion started: 8 October 2020 c Author(s) 2020. CC BY 4.0 License.

Figure 2:
Regional geology map of the study area (adapted from Bamousa et al., 2017).

460
Altitude values (x) are not provided with a numbered scale for unknown coefficient (k) value in the following equation: x = 2kπ. If k for example is equal to 1, the lowest altitude value is equal to 2π. Therefore, the interval of the provided range is equal to 50 m/colour, ranging from 400 m to 1,000 m according to the topographic map.